1. Field of the Invention
This invention relates to communication networks. More specifically, the present invention relates to communication over optical networks.
2. Background of the Invention
Various topologies can be used in a network. One such network topology is a ring. Different types of transport technologies can be used on a ring network. One class of these transport technologies relies on multiplexing (e.g., time division multiplexing (TDM), wave division multiplexing (WDM), dense wave division multiplexing (DWDM), etc.).
An optical standard such as Synchronous Optical Network (SONET) or Synchronous Digital Hierarchy (SDH) in conjunction with a multiplexing scheme is used to deliver various rates of traffic over high-speed optical fiber. SONET/SDH is a transmission standard for optical networks that corresponds to the physical layer of the open standards institutes (OSI) network model. One of the protection schemes for SONET/SDH in a ring network involves automatic protection switching (APS) in a bi-directional line switched ring (BLSR) architecture. There are different types of BLSR, including two fiber and four fiber. In addition to BLSR, there is also a unidirectional path switched ring (UPSR) architecture.
FIG. 1 illustrates a type of multiplexing ring network element and the typical network architectures within which it is used according to the prior art. In particular, FIG. 1 illustrates the multiplexing ring network elements 110, 150 and 160. Each of these ring network elements includes a cross-connect (114, 154 and 164, respectively). Each of these cross-connects is connected to a number of line card slots within the chassis of the network element. It should be noted that the different slots within a network element are connected to the cross connect with either a low-speed or a high-speed bandwidth connection. In particular, there are at most 4 line card slots connected to the cross-connect with high-speed bandwidth connections (each of the rest of the line card slots are connected to the cross-connect with low-speed bandwidth connections). The insertion of the appropriate line card into one of these slot forms an interface. For instance, the cross-connect 114 of the network element 110 is connected to each of the linear interfaces 112A through 112I with a low speed connection, while the cross-connect 114 is connected to each of the ring up to 4 interfaces 116A–116D with a high-speed connection. The network elements are designed this way because the low speed interfaces are used for providing linear connections, whereas the high-speed interfaces are used for connecting the network element to a ring.
Certain network elements used for the purposes shown in FIG. 1 have a simple add-drop MUX instead of a cross-connect. However, this design suffers from a well-known fragmentation problem regarding the time slots on the rings. This fragmentation problem was the motivation to include a cross-connect in the network elements of FIG. 1. Specifically, the cross-connect of FIG. 1 is used to shift traffic received by a network element on certain incoming time slots of the ring to different outgoing time slots of that ring (referred to as time slot interchanges) to reduce or eliminate fragmentation.
These network elements are typically used in two different types of network architectures: 1) an access network architecture; and 2) a hubbed network office architecture. In the example of FIG. 1, the network element 110 is being used in an access network architecture. Specifically, each of the linear interfaces 112A–112I is connected to one or more pieces of customer premise equipment 100A–100K (e.g., using one or a combination of T1, T3, DS1, DS3, etc.). In contrast, the ring interfaces 116A–116D are connected to a ring 145 that includes a network element 140 and the network element 150.
The ring 145 is often referred to as a trunk or collector ring and consists of either two or four fibers. The linear connections to the customer premise equipment are referred to as tributaries of the collector ring. The tributaries are used to add and drop traffic between the ring 145 and the different pieces of customer premise equipment. The sum of the bandwidth to the linear interfaces typically does not exceed the sum of the bandwidth to the ring interfaces.
This access network architecture is typically used in a metro setting. In particular, the pieces of customer premise equipment 100A–100K typically reside in different office buildings. This customer premise equipment provides metro access to a metro collector ring (e.g., the ring 145). While the ring 145 has significantly greater bandwidth than any of the connections to the customer premise equipment, the sum of the bandwidth required for the actual traffic to and from the customer premise equipment cannot exceed the bandwidth available on the ring 145. This is why the sum of the bandwidth connecting the cross-connect 114 to the linear interfaces 112A–112I typically does not exceed the sum of the bandwidth from the cross-connect 114 to the ring interfaces 116A–116D.
In the example of FIG. 1, the network elements 150 and 160 are being used in a hubbed network office architecture. A hubbed network office is a location where multiple collector rings are interconnected. The network elements 150 and 160 have the same architecture as the network element 110. Specifically, the network elements 150 and 160 have ring interfaces 156A–156D and 166A–166D respectively connecting the network elements to the rings 145 and 165. These ring interfaces have a high-speed bandwidth connection to their respective cross-connects. In addition, the network elements 150 and 160 respectively include linear interfaces 152A–152I and 162A–162I that are each connected to their respective cross-connects with low bandwidth connections.
The rings 145 and 165 are interconnected by one or more linear connections between the network elements 150 and 160 using the linear interfaces 152A–152I and 162A–162I. This interconnection allows for the passing of traffic between the rings. For example, traffic from the ring 145 may be “dropped” from ring interfaces 156A–156D to tributary interfaces 152A–152I, and then “added” from tributary interfaces 166A–166D to the ring 165.
The architecture of the network elements and the network architectures in FIG. 1 suffer from several limitations. With respect to the access network architecture, it is relevant to understand the manner in which the linear connections are selected, deployed, and provisioned. Specifically, when connecting a piece of customer premise equipment to the ring 145, a selection must be made as to what amount of bandwidth connection should be deployed. This decision is based upon an estimate of the current and future bandwidth needs. For the network operator with many customers, a per customer decision must be made for how much each customer or area will grow. Once the prediction is made, the connection must be deployed (e.g., by tearing up the road between the network element 110 and the customer premise equipment). Once the connection to the customer premise is deployed and provisioned, the network operator begins to collect revenue based upon the bandwidth of that line.
When one of the above predictions is wrong and/or enough time has passed, it is not uncommon for a higher bandwidth connection to need to be deployed. When this happens, the basic steps listed above are again performed. This process is expensive and time consuming for several reasons. Specifically, it takes a long time and it is expensive to re-deploy a higher bandwidth line (e.g., digging up the road again). In addition, the increased price for the higher bandwidth connection cannot be charged by the network owner until the higher bandwidth connection is established. Thus, the time it takes to deploy the line reduces the revenue for the network owner. Effectively, a less expensive network element (due to the lower speed bandwidth connections between the linear interfaces and the cross-connect, as well as the less complex cross-connect required as a result) has resulted in a greater operating cost (e.g., the cost of deploying higher bandwidth connections, as well as the loss of the higher billing rate for the increased bandwidth connections while they are being installed).
With respect to the hubbed network office, a separate network element is required for each ring. Each of these network elements takes up expensive rack space and power within the hubbed network office. In addition, connecting two rings requires the installation of cross-connects in each of the network elements, as well as one or more linear connections between the two network elements. The establishment of these network connections requires expensive network operator time and is subject to human error. Furthermore, the service provider most predict how much bandwidth must be exchanged by each pair of rings, and hardware linear tributaries accordingly. Changes in requirements require re-wiring the interconnects between the network elements, possibly: 1) disturbing traffic being carried on the existing interconnects; and 2) requiring the addition of more network elements or rings in order to provide the additional interconnects. On a network-wide basis, these complications lead to circuit provisioning delays of many months.
Additional limitations with respect to the hubbed network office can be understood with respect to FIGS. 9A–B. In particular, FIGS. 9A–B illustrate two different techniques for interconnecting more than two rings in a hubbed network office. Both of FIGS. 9A and 9B illustrate a network element 905, a network element 910, a network element 915, and a network element 920. A different ring is connected to each of these network elements. For purposes of illustration, one of these rings (ring 900) is shown to include four fibers, whereas the remainder of the rings are shown to include two fibers. It should be noted that this is done merely for exemplary purposes, and thus, any one of the rings may be a two or four fiber ring.
FIG. 9A illustrates a hubbed network office in which the network elements are interconnected by linear interfaces according to the prior art. In particular, FIG. 9A shows that each of the network elements is connected to the other network elements via one or more linear interfaces. In both of FIGS. 9A and 9B, a hash through a line indicates that that line represents one or more connections. The hashes on the lines interconnecting network element 905 to the network elements 910, 920, and 915 are respectively labeled R, S and T. This is done for the network element 905 for purposes of the following illustration. Since the aggregate bandwidth of the linear interfaces of any one of the network element 905, 910, 915, and 920 is no greater than the aggregate bandwidth of the ring interfaces, a limited amount of bandwidth can traverse between any given set of two rings. For example, where the network element 905 has X linear interfaces, T+S+R must be less than or equal to X. In addition, the bandwidth of the X linear interfaces is no greater than the bandwidth of the ring 900. Thus, if it was desired to move the traffic from ring 900 to the ring connected to network element 915, all of the linear interfaces of the network element 905 would be required for that purpose.
It is also relevant to note that the connections between each of the network elements are hardwired, and thus, incorrect estimates in the amount of bandwidth that must be provisioned between these network elements requires rewiring the interconnects.
FIG. 9A also illustrates that a digital cross-connect 930 may optionally be included. This digital cross-connect 930 could be connected to each of the network elements to allow for greater flexibility in switching and/or the adding/dropping of other traffic. It should be noted that the number of ports required on the digital cross-connect 930 is driven by the amount of bandwidth with which that digital cross-connect is going to be connected to each of the network elements in FIG. 9A.
FIG. 9B illustrates a hubbed network office in which the network elements are interconnected via a digital cross-connect according to the prior art. In particular, FIG. 9B illustrates that each of the network elements is connected to a digital cross-connect 935 by their linear interfaces. While the digital cross-connect 935 allows for a greater flexibility in switching between the rings, it should be understood that the digital cross-connect 935 requires a significantly larger number of ports for all of the connections to the linear interfaces of each of the network elements. Due to the cost of a digital cross-connect with this many ports, the scheme illustrated in FIG. 9B is rarely used.